Spaces:
Running
Running
Update models.py
Browse files
models.py
CHANGED
@@ -4,56 +4,18 @@ import torchvision
|
|
4 |
import torchvision.transforms as T
|
5 |
from PIL import Image
|
6 |
|
7 |
-
class
|
8 |
-
def __init__(self,
|
9 |
-
super(
|
|
|
10 |
|
11 |
-
self.
|
12 |
-
|
13 |
-
# Workaround to bypass HTTP Error 403 rate limit exceeded
|
14 |
-
torch.hub._validate_not_a_forked_repo=lambda a,b,c: True
|
15 |
-
self.model = torch.hub.load("facebookresearch/dinov2", "dinov2_vitb14_lc")
|
16 |
-
|
17 |
-
self.model.linear_head = nn.Sequential(
|
18 |
-
nn.Linear(3840, 512, bias=True),
|
19 |
nn.ReLU(),
|
20 |
nn.Linear(512, 256, bias=True),
|
21 |
nn.ReLU(),
|
22 |
-
nn.Linear(256,
|
23 |
)
|
24 |
|
25 |
-
self.model.to(self.device)
|
26 |
-
|
27 |
-
self.transform_image = T.Compose([
|
28 |
-
T.Resize((224, 224)),
|
29 |
-
T.ToTensor(),
|
30 |
-
T.Normalize(mean=[0.485, 0.456, 0.406],
|
31 |
-
std=[0.229, 0.224, 0.225])
|
32 |
-
])
|
33 |
-
|
34 |
-
self.model_name = "dinov2"
|
35 |
-
|
36 |
-
def load_image_from_filepath(self, img: str) -> torch.Tensor:
|
37 |
-
"""
|
38 |
-
Load an image as filepath and return a tensor that can be used as an input to model.
|
39 |
-
"""
|
40 |
-
img = Image.open(img).convert('RGB')
|
41 |
-
|
42 |
-
transformed_img = self.transform_image(img)[:3].unsqueeze(0).to(self.device)
|
43 |
-
|
44 |
-
return transformed_img
|
45 |
-
|
46 |
-
def load_image_from_pillowimage(self, img: Image.Image) -> torch.Tensor:
|
47 |
-
"""
|
48 |
-
Load an image as Pillow Image and return a tensor that can be used as an input to model.
|
49 |
-
"""
|
50 |
-
transformed_img = self.transform_image(img)[:3].unsqueeze(0).to(self.device)
|
51 |
-
|
52 |
-
return transformed_img
|
53 |
-
|
54 |
def forward(self, x):
|
55 |
-
|
56 |
-
x = self.load_image_from_filepath(x)
|
57 |
-
if isinstance(x, Image.Image):
|
58 |
-
x = self.load_image_from_pillowimage(x)
|
59 |
-
return self.model(x)
|
|
|
4 |
import torchvision.transforms as T
|
5 |
from PIL import Image
|
6 |
|
7 |
+
class LinearClassifier(torch.nn.Module):
|
8 |
+
def __init__(self, input_dim, output_dim):
|
9 |
+
super(LinearClassifier, self).__init__()
|
10 |
+
num_classes = len(index_to_species.keys())
|
11 |
|
12 |
+
self.linear_head = nn.Sequential(
|
13 |
+
nn.Linear(input_dim, 512, bias=True),
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
nn.ReLU(),
|
15 |
nn.Linear(512, 256, bias=True),
|
16 |
nn.ReLU(),
|
17 |
+
nn.Linear(256, output_dim, bias=True)
|
18 |
)
|
19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
def forward(self, x):
|
21 |
+
return self.linear_head(x)
|
|
|
|
|
|
|
|